| Abstract [eng] |
The process of morphological characterization of a specimen using scanning electron microscopy (SEM) generally produces a magnified specimen image, which is built out of the ejected electron beam raster scan data and the initial position of the beam. The scanning electron microscope provides large depth-of-focus and high-contrast images at a large variability in magnification. Nevertheless, the resulting image is literally a two-dimensional intensity distribution array and usually does not carry any additional specimen-related metadata, except for the observational circumstances. Extracting spatial data from the two-dimensional images can be difficult, so to achieve this, the specimen must be handled in a special way and, also, the produced micrographs must be digitally processed afterwards [1-4]. In materials science, various techniques for three-dimensional reconstruction of microstructures have been applied successfully for decades, such as X-ray and electron tomography [5-8]. X-ray computed micro tomography, sometimes combined with micro-X-ray fluorescence technique is often used in medicine and material characterization as a non-invasive method for visualizing the three-dimensional structure of various objects [9]. It was demonstrated, that optical coherence tomography allows rapid evaluation of coating or skin surface morphological properties and three-dimensional images could be reconstructed [10, 11]. The unique capabilities of ultra-short pulse femtosecond lasers have been integrated with a focused ion beam platform to create a new system for rapid 3D materials analysis [12]. Consequently, the combination of integrated focused ion beam-scanning electron microscope (FIB-SEM) provides serial sectioning capability in combination with imaging technique. Image analysis has been recently used for quantitative threedimensional characterization of different materials [13-17]. Therefore, the improvements in the detection of microstructure of different materials open the 7 new areas of applications and provide a strong motivation for studying electron microscopy data processing possibilities in general. Electron microscopy has made great impact on structure determination, however, only TEM showed a strong potential for structural studies of nanocrystals, interfaces, phase boundaries and their surfaces [18-20]. On the other hand, SEM stereoscopic technique was also used to determine the threedimensional surface structures [21, 22]. It is very well known, that stereoscopy simulates human vision by combining two pictures of a specimen from two slightly different angles. The depth of the produced stereoscopic images depends on the chosen angle where 2.5, 5 or 10 degrees [23] is usually enough to produce a sufficient depth perception to the viewer. The success of this technique highly depends on the microscope type and its instrumental precision. Stereo photography requires precise compucentric tilting of the observed sample, so that during the tilt, image focal points are preserved in the microscope viewport [24]. This is difficult to achieve for some SEM models, where the accuracy is linearly decreased, as the magnification increases. However, if the tilt is precisely compucentric, anaglyphs of magnification up to 50000 times can be easily produced. Otherwise, the SEM operator must manually apply the compensation of the shift that happens during the tilt. Thus, until now the suitable models for the extrapolation of three-dimensional sample data out of the SEM images are lacking, and consequently, the precise qualitative and quantitative microstructural analysis is still an open question. The motivation of this work was to develop new SEM data processing model on standard materials and verify this model on compounds with unknown microstructure. This is the main novelty of PhD thesis. The development of the program model for the application of 3D reconstruction technique was also in the field of interest of this doctoral dissertation. 8 The main objective of this work was to develop an advanced scanning electron microscopy data processing model for the extrapolation of threedimensional sample data out of the SEM images. For this reason, there were formulated tasks as follows: - To propose advanced model how to extrapolate, measure and interpret 3-dimensional sample data out of the SEM images of gold nanoparticles, which have been selected as model material for the spatial 3D surface reconstruction. - To verify the developed 3D reconstruction technique by investigating medium resolution aluminium-tungsten dendrites (with various spacing in the dendritic structure) and p-type monocrystalline Si (100) textured wafers with thickness of 200 m. - To elaborate a simple, inexpensive, environmentally benign, morphology and crystallinity-semicontrolled sol-gel synthetic approach for the preparation of calcium hydroxyapatite (Ca10(PO4)6(OH)2, CHAp). - To apply the proposed 3D reconstruction technique to unknown nanostructures of CHAp. - To develop the program model for the application of 3D reconstruction technique for any SEM images. |
